Ermene

decision intelligence

Patent filed UIBM · 04.05.2026
Scroll

How much of the world
enters your decisions?

The risk isn't ignoring the outside world. It's watching it from a distance.

Ermene redraws the limits of observation.

It gathers your data, the market, the internet, the signals in circulation. It assigns each one its weight, its volatility, its persistence.

It fuses horizons. It doesn't hand you a point of view: it listens, understands what you're looking for, and the more you talk, the deeper it goes.

How it's built

Five separate layers. One single line of reasoning.

No layer talks directly to the next. None trusts without verification. Separation is the guarantee.

01 Data Ingestion & Classification

The roots of the world.

What do you really need, to understand your world?

What you already know — your internal data, the CRM, sales, costs. What the market is saying — competitors, trade fairs, prices, sector movements. What's happening outside — social, magazines, forums, distant trends that touch you closely.

The first step is yours. Together we define the sources.

Then the system collects, verifies, cleans, structures. Every signal receives a three-dimensional identity that determines its weight — and that is updated over time.

In practice

  • Multi-source ingestion
  • V Volatility
  • N Nature
  • A Authority
  • Dynamic registry
02 Signal Computation

From data to signal.

How would you describe your world?

Not with a number. Not with a chart. With the connections that exist between things. Like a brain: what counts isn't the single neuron, but how it fires in relation to the others.

Ermene observes those connections. Measures them. Classified data becomes signal; the system watches how each one behaves over time.

The KPIs you usually see at the end of an analysis appear here at the start. They aren't the result. They're building blocks.

In practice

  • Non-linear pattern recognition
  • Cross-domain correlations
  • Traceable to source
  • No black box
03 Signal Integration

The generative transformation.

Synapses transform.

It isn't the sum of the data. It isn't an average of the signals. It's something new that emerges when everything connects.

From A and B you don't get A+B. You get C — something no single source could ever contain on its own.

Your data stays at the centre: it's the truest you have. But nothing is shut out. Not even the noise of the forums. The result is something alive, that changes with the world.

In practice

  • Quality-gated
  • Temporal-aware
  • Differentiated decay
  • Non-additive synthesis
04 Validation & Calibration

The world, verified and alive.

Should I take it on faith? — No.

Every prediction is checked against what actually happened. Independently, separately, autonomously. Validation keeps the model aligned with reality.

The world doesn't stand still. What is true today may not be true tomorrow. The model Ermene has built is not static: it adapts, it evolves. Ermene is alive.

Not faith. Evidence.

In practice

  • Immutable T0 snapshots
  • T+1 comparison
  • No feedback loop
  • Source calibration
05 Hermeneutic Interface

One word from you.

The system is waiting. Tell it what's on your mind.

Language is how we connect to the world. To people. To the decisions that matter. Ermene was born, built and generated to answer that.

Numbers, charts, analyses don't disappear — they change role. They become part of the language with which Ermene answers you.

The more you know each other, the sharper the understanding. You become part of Ermene too. Ermene grows with you.

In practice

  • Natural conversation
  • Read-only between L3 and L4
  • Distinct epistemes
  • Dialogue memory
Technology

Not an idea. A system.

Verifiable against the state of the art, and protected by a filed patent.

Dimension BI & Dashboards Tableau, Power BI General-purpose LLMs ChatGPT, Claude RAG / Search Perplexity, Glean Ermene
Sources Internal structured data only Pre-training corpus + web Documents + indexed web Internal + market + external, classified with V·N·A
Weighting Static, manual Implicit in model weights Relevance score Three-dimensional identity empirically verified over time
Transformation Aggregation, sum, average Probabilistic generation Retrieval + generation Non-additive generative. A + B → C
Validation Manual, ex-post None on individual answers Citation-based T0 snapshot → T+1 comparison, independent and traceable
Interface Graphical, click-driven Generic chat Q&A on documents Hermeneutic. Dialogue memory, shared horizon
IP protection Patent filed UIBM · 8 claims

Artificial intelligence architecture system for the construction, management and querying of decision models.

A dual-generative architecture of stacked physical layers, operating across separate, complementary epistemic planes. Eight claims — generic and inclusive — covering the end-to-end process.

Application number
102026000012436
Filing date
04 / 05 / 2026
Classification
G06N20
Status
Filed · under examination

Coverage

🇮🇹
Italy UIBM · national application
🇪🇺
EPO extension Planned within 12 months
🌐
PCT Under evaluation
Where the method comes from

Truth and Method

Hans-Georg Gadamer · 1960

«Not through neutral observation, which doesn't exist. But through dialogue.»

Every interpreter carries a horizon: their own context, their own history, their own questions. When they meet something external — a text, an event, a signal — they don't receive it passively. They fuse it with their own horizon.

From that fusion a new understanding is born. Something that didn't exist in the interpreter alone, nor in the text alone. A third element — situated, contextual, unrepeatable meaning.

And then Gadamer adds something even deeper: the process doesn't end. The new understanding becomes part of the interpreter's horizon, who then meets the next experience with different eyes. The hermeneutic circle never closes. It refines itself, continuously.

Language sits at the heart of all this. For Gadamer it isn't a tool we use to transmit thoughts already formed. It is the place where understanding happens. Before language, there is no understanding yet — only raw experience.

Dialogue — real conversation, the kind in which both sides change — is the highest form of understanding we know.

The instruments change.
The way we understand the world — doesn't.

See it at work

Ermene for Tennis.

Our technology, applied.

111K Events analysed
25 Years · ATP / WTA
5 Separate layers

A Sequence Risk Engine that doesn't evaluate the tournament — it evaluates the sequence: the gap from one tournament to the next, the load that builds up, the surface that changes, the moment in the season.

Source calibration with mandatory human approval before any change. Not a forecast: a structured assessment you can measure, verify, contest.

Empirically validated on 25 years of ATP/WTA data.

You can start here

  • “How would you change the ATP tour to make it less risky for the players?”
  • “What's the risk for tall players?”
  • “Compare Sinner and Alcaraz.”
  • “How should Badosa set up her calendar to balance points and risk?”
While you try it

Think of your world.
What would you ask?

Let's talk.

If this is the kind of instrument you'd want beside your decisions, write to us. A real person replies.

info@ermene.it